changes of the root. For instance, in the past tense, we have generally, with the exception of the labial consonants, for the first consonant a surd; in the present, the corresponding aspirate or sonant; and, in the imperative, the corresponding aspirate. If, further, the past tense has the vowel a, then we have frequently e for the vowel of the present, and o for that of the imperative; thus perf. b-kas, pres. h-ges, imperat. khos v.t. 'to split.' If, however, the root-vowel is i or u, then it remains unchanged in all the tenses, as, for instance, past b-kum, pres. h-gum, imperat. khum v.t. 'to kill.' Again, intransitive verbs are usually distinguished from transitive by the aspiration or weakening of the surd of the perf. tense of the transitive verb, as the above-mentioned perf. b-kas v.t. 'to split'; but perf. gas, pres. h-gas v.i. 'to split'; there being no change of the initial consonant or vowel change in the intransitive form. If, however, a verb commences with a labial consonant, we have an aspirated surd for the transitive, and an unaspirated sonant for the intransitive form, as perf. phral v.t. 'to separate'; perf. bral, pres. h-bral v.i. 'to separate.' Burmese, on the other hand, not only in regard to labials, but throughout, forms transitives from intransitives by aspirating the initial consonant, as kya 'to fall down,' khya 'to throw down,' and so on with many others. In Bahing, again, the intransitive form has a sonant for the initial consonant, the transitive a surd, as gikko 'be born,' kikko 'beget'; bokko 'get up,' pokko 'raise up.' In some languages a modification of the pronominal suffix of the verb takes place in order to change it from the subject into the object. Thus, in Bahing, khi-ka-mi 'we quarrel with them,' khi-ki-mi 'they quarrel with us.' In Khyeng, to form the negative voice, besides the insertion of a letter, an initial surd is changed into a sonant, sit-phá-u 'may go,' zit-pha-mbu 'may not go.' But, as I have said, the few sketches of grammar we have are in general so short that we cannot expect to find in them any account of root- changes when they are merely exceptions to the ordinary rules.
But though the group itself is a very vague one, many of the languages may easily be subdivided into classes on
